Comment by xstartup | Hacker News Reader

Comment by xstartup | Hacker News Reader

As someone who has multiple startups.

Here are the main benefits of patents:

1. Helps in IPR negotiations. When multiple startups are competing in same space and poaching same employees, highly likely your tech will end up being similar to your competitor. Maybe some of the processes are exact copies. Then if you've no patents and you get sued, good luck defending yourself.

We solve it by creating landmines of patents around our actual application. So, that if we end up infringing someone else's patent, we'll be able to find some of their infringement then negotiate the settlement.

If you've nothing, you can't defend and your settlement will not be something in your favor.

Conversation with competitor: Mate! you violated my patent X and I violated your patent Y, we'll get nothing out of the legal battle, the damage is of the same amount. Why not stay quiet and keep the lights on? Who doesn't love peace?

2. Patents in employee's name often boost their confidence (if not their wealth). So, you get some productivity boost from their motivation. Sometimes, it's like keeping a score for them and we payout enough to host a memorable dinner with their family.

Conversation with Self: You know like I've 15 patents under my belt already and 5 more pending. I am an expert already I think. Everyone will be proud of me.

3. Keeping new entrants out.

Conversation with Self: Oh, it will be expensive to navigate this patent landmine, so let's choose some other niche/industry for my new startup idea.

Thank you for summing up all the problems patents cause in a market. You just made the point why we must get rid of patents ASAP :)

I doubt they will every go away. You have to make peace with that.

You can't get rid of patents by arguing that they distort markets. That's the entire point of them in the first place. To incentive people to create new things by restricting competition for a limited period of time.

You conveniently forget the final part. Patents give you a limited monopoly in exchange for full disclosure of your invention. By filing for a patent you accept that all the knowledge it contains becomes public domain. That’s the original deal. The comment I reacted on describes the IMHO perverted way patents are used nowadays. They don’t share knowledge (especially software patents), they only incentivise defensive, litigious behaviour in the market.

They don’t share knowledge (especially software patents)

Can you explain how they do not share knowledge?

They only incentivise defensive, litigious behaviour in the market.

1. A patent grants you a right

2. If someone threatens your right, you are going to defend it. Or are you suggesting that we should not use legal methods to defend our right? Any right? Or just patent rights is the exception you propose?

3. If you threaten other's right and they threaten yours, you have a way to negotiate at a table outside the court and save society's precious resources.

I still don't see how it's bad. Please enlighten me.

>You can't get rid of patents by arguing that they distort markets.

You make a good point here that patents are intended to distort markets. However they are intended to distort markets in particular ways and the question is, do they distort markets in the ways in which they were intended. Specifically do they reward and encourage investments in innovation for the public good? Answering this question in general is very tricky, but within the cryptography space it appears that they do not. There are several reasons for this that I am aware of and probably the few that I am missing:

1. Companies and open source projects avoid using patented cryptography. This means that making money off of patented cryptography is hard and the public does not benefit from these innovations until the patent expires.

2. Researchers generally boycott patented algorithms, for instance submitting algorithms for government crypto contests like SHA-3/AES often require ceding all patent rights. Patented algorithms are less well studied and aren't standardized so don't benefit from the research or operational community.

3. Most cryptographic algorithms require at least two parties. Patents typically require both communicating parties to have a license. Thus network effects favor open standards unencumbered by patents.

4. The US government has a long history of not paying for patented cryptographic inventions due to the vagaries of US law. Thus, patents provide only minimal protection against historically one of the biggest buyers of cryptographic inventions, the US government.

5. The public good of having cryptographic algorithms be unencumbered by patents is so great that people will just figure out how to get around your patent. This is what happened with Schnorr signatures. The US government altered the Schnorr signature algorithm just enough to create an algorithm, DSA (Digital Signature Algorithm), such that the patent didn't apply. DSA isn't quite as good as Schnorr, but DSA is better than Schnorr+patent so nearly everyone used DSA until the patent expired.

Patents in the context of cryptographic algorithms appear to not help much in terms of encouraging and rewarding innovation and they also appear to act counter to the public good.

Many of the same arguments apply to patented algorithms in image processing and computer vision as well. SIFT and SURF are probably the two most famous examples, e.g. [0] and [1].

Two quick anecdotes about this. The result of the message board thread of [0] once led to a compliance team in my former company to ban all use of OpenCV for computer vision applications (which, obviously, is insane). They took an exceedingly risk averse stance that if any of our software imported a third party OpenCV library that contained SIFT, this potentially opened us up to a lawsuit, even if we did not actually use SIFT (and might incur costs for us to prove we didn't use SIFT). Needless to say, this created an unbelievable amount of pain for being on computer vision and image processing teams there. Lots of re-implementing of basic algorithms, but without the benefit of years of community development, testing, etc.

And [1] below indicates the sort of patent-avoiding contortions people will go through in research. If your algorithm is patented, typically it means people won't use it and will do whatever is needed to avoid it, even just abandoning a certain line of research all together and choosing something else without the possible patent hurdle.

In the most innocuous cases, it just means the patented algorithm receives far less research attention and becomes irrelevant faster (giving the interpretation that the patent acts as a very short term royalty grab by the patent author, which seems antithetical to the stated public innovation goals of a patent system in the first place).

In worse cases, the patent acts like a research roadblock that can thwart useful lines of research from others. When patents allow for uncompensated academic work to use the algorithm royalty-free, that can sometimes be an OK middle ground, but still doesn't seem to be in the spirit of the whole idea to spark and catalyze innovation in the first place.

[0]: < http://opencv-users.1802565.n2.nabble.com/SURF-protected-by-... >

[1]: < https://dsp.stackexchange.com/questions/1288/what-are-some-f... >.

I suspect that fields sufficiently similar to Mathematics are a poor fit for the patents and will inevitably adopt the norms of Mathematics research.

This is basically the story of why ORB-SLAM didn't become the de facto commercial slam implementation. They wanted iirc 30k per license, which is impossible for any startup.

The point of patents is to spur innovation. This new role is one they are wholly unsuited for.

I have a patent on something completely obvious and stupid, filed on my behalf by a company I used to work for. It was based on some offhand remark I made at a meeting. Every engineer at the meeting is on the patent, including the VP of the company who was not even present-- and I am last.

I do not remember making the remark, but my colleagues who all took credit for it, assure me I did.

+ I do think patents prevent a race to the bottom.

Imagine if everyone had the same process, the only distinction would be the amount of output you can get per employee (as the employee compensation is one of the biggest cost) or marketing/brand (another big expense category) or customer service. Companies might end up exploiting employees or blatantly misleading their customers in order to attract the business.

First two of which are detrimental to society at large and not the optimum uses of the available resources.

Third of which, the improved customer service is a plus. But the user perception of better customer service is not exactly better customer service. But yea incumbents are often not willing to provide better customer service when there is a moat (guaranteed value), I've experienced this first hand with companies like Facebook Ads, Google Cloud etc.... some of them have improved now while some still suck.

I do not think patent stop disruptions from happening.

If you create a new process which is vastly superior to the old process without infringing existing patents, you can do that.

If you improve a patented process, you can sell it to the company owns the patent but they might not pay you the best price as they have the monopoly and they do not need to.

We need some mechanism (law?) to facilitate a fair payment for an improvement over an existing patent to avoid monopolistic exploitative practices. If an organization can truly benefit from an improvement and is in position to make a fair valuation and payment, IMHO it should.

Reply on news.ycombinator.com